Regional variation in medical classification agreement: benchmarking the coding gap.

The growing use of classification and coding of patient data in medical information systems has resulted in increased dependence on the accuracy of coding practices. Information maintained on systems must be trusted by both providers and managers in order to serve as a viable tool for the delivery of healthcare in an evidence-based environment. A national survey of health information managers was employed here to assess observed levels of coder agreement with physician code selections used in classifying patient data. Findings from this survey suggest that, on a national level, the quality of coded data may suffer as a result of disagreement or inconsistent coding within healthcare provider organizations, in an era where physicians are increasingly called upon to enter and classify patient data via computerized medical records. Nineteen percent of respondents report that coder-physician classification disagreement occurred on more than 5% of all patient encounters. In some cases disagreement occurs in 20% or more instances of code selection. This phenomenon occurred to varying degrees across regions and market areas, suggesting a confounding influence when coded data is aggregated for comparative purposes. In an evidence-based healthcare environment, coded data often serves as a representation of clinical performance. Given the increasing complexity of medical information classification systems, reliance on such data may pose a risk for both practitioners and managers without consistent agreement on coding practices and procedures.